Learning is an essential function of the nervous system that allows animals to modulate behavior with adaptive values. While increasing amount of knowledge on the molecular underpinnings of learning provide insights into the mechanisms underlying learning, our understanding cannot explain behavioral changes in most learning paradigms. One major challenge of the field is to link the function of the underlying neuronal network with behavior and to address how the property of neural circuitry encodes learning. We use the genetic model organism C. elegans to address this question. In the past funding period, we have established a form of aversive olfactory learning whereby the nematode learns to avoid the smell of pathogenic bacteria that make it ill. This form of learning is analogous to Garcia effect, in which animals learn to avoid the smell or taste of a food that is associated with stomach distress. Using this learning paradigm, we have characterized the structure and function of the underlying neuronal network. Particularly, we show that a serotonergic neural circuit composed of the serotonergic neuron ADF and the downstream interneuron RIA, as well as motor neurons specifically regulate learned olfactory preference. The serotonin signal in ADF regulates the aversive learning on pathogenic bacteria. ADF responds to bacterial odors with increased intracellular calcium signals and the C. elegans homolog of CaMKII, UNC-43, acts in ADF to regulate learning. The postsynaptic neuron RIA is critically required for the aversive learning. RIA displays compartmentalized axonal activity that is correlated with head movement. Meanwhile, RIA axonal compartments also display synchronous activity that is evoked by olfactory stimuli. Interestingly, we show that the aversive training modulates the activity pattern of ADF and RIA in a way that is consistent with training-induced behavioral changes in olfactory preference. Thus, we hypothesize that these learning-correlated changes in the functional attributes of ADF and RIA neurons encode learning. We propose to test this hypothesis by characterizing the regulatory mechanisms and function of the learning correlates in ADF and RIA. We will first define the interaction between these two learning correlates by testing the possibility that the learning correlate in ADF regulates the learning correlate in RIA. We will als characterize the regulation of these learning correlates by examining the effect of several genetic factors that we have identified to mediate learning. We will also define the neurotransmission of ADF that regulates RIA activity. Second, we will characterize the function of the learning correlates in ADF and RIA. We will use molecular and optogenetics to manipulate the property of these neurons to (1) eliminate the training-induced changes in their activity patterns; and (2) build the learning correlates with genetic methods in the key neurons of the circuit, and then test the resulting effects on olfactory learning. These studies will revea how experience modulates the function of a neural network and leads to experience-dependent behavioral changes.